This invention relates to systems for stabilizing video images and in particular to systems for use in telecines.
Telecines suffer from image stability problems. A film is transported through the telecine's optical system by a complicated series of mechanical devices which use the sprocket holes in the film to guide it through a predetermined path of travel. As a film is used the sprocket holes will be subjected to wear and tear thus making it impossible to use them accurately to position every frame of the film in the telecine.
Furthermore, if the film passing through the telecine is a print from an original a variable error in the position of the actual image relative to the film edges will be present as a result of the process by which copies of the film are made. The sprocket holes in the original would be subject to wear and tear and the error in positioning the original when making a copy will contribute to this variable error. In such a case a telecine which had error correction to produce 100% image stability would still produce an unsteady video image.
Motion vector estimation techniques are well known in video systems and are used to identify the motion of a whole video frame or a portion thereof relative to its immediate predecessor. In a telecine this information can be used to correct the position of the current frame of the video signal relative to its predecessor thereby producing a stable output image. Preferably a motion vector estimator is used which has sub-pixel accuracy since even very small positional errors can be noticed by the human eye in certain situations, for example, when electronically generated titles are overlaid on a transferred sequence these are completely steady with respect to the image edges.
It is sometimes desired to change the shape and size of the images in a sequence in situations such as when two pieces of film material are to be joined by a matte process. The alteration to the shape and size can be achieved by use of a digital image processor after scanning of the film by the telecine. However, the effectiveness of this is limited by the resolution of the original sampled images. This can cause problems with image rotation and image zooming and resultant images are likely to be either very "soft" or "pixellated".
The Rank Cintel URSA telecine incorporates a scanning mechanism which allows special effects such as image rotation and image zooming to be carried out before a digital representation of the image is produced, thereby maintaining image resolution. This is achieved by altering the shape of the flying spot scan to that of the desired effect. For example, an image zoom is accomplished by scanning the same number of lines in a reduced area of the film. Image pans, rotations, and other effects such as image stretching can also be performed in this way.
Using the known technique of Producing motion vectors between successive frames of the image to correct for film instability does not work when such a zoom, pan, or other special effect has been included and has altered the shape of the flying spot scan. This is because the motion vector estimation technique tracks the global motion of the image from frame to frame and would be unable to separate out motion due to scanning effects from motion due to film instability.